Predicting retinal haemorrhage following retinal vein occlusion

Bhattacharya, Atrayee (2026) Predicting retinal haemorrhage following retinal vein occlusion. PhD thesis, University of Glasgow.

Abstract

The retina is the layer of sensory tissue that lines the interior surface of the human eye. The retinal cells are nourished by a network of blood vessels, which are fed by a single artery/vein pair that enters and exits through the optic nerve. Retinal vein occlusion is a blockage in one of the veins of the retinal circulation, which can subsequently lead to retinal haemorrhage. Such blockages often occur at points of arterio-venous crossing, where the retinal vein and artery share a common sheath, and gradual swelling of the artery (due to arteriosclerosis) results in compression of the adjacent vein, narrowing the blood vessels, damaging their wall and eventually leading to a venous thrombus which rapidly occludes the vein. Retinal vein occlusion is the second most common retinal vascular disease and a common cause of vision loss in older patients. For the first time, we aim to understand how occlusion in one part of the retinal circulation can lead to haemorrhage in other parts of the network using a combination of cutting-edge image analysis and a mathematical model based on continuum mechanics. The clinical images suggest that the bursting of vessels occurs several generations upstream of the site of arterio-venous crossing. To investigate this clinical observation, we develop a one-dimensional network model composed of three generations of vessels informed by clinical images; we consider a locally applied (external) perturbation to drive an occlusion of the vein. We find that wave propagation only plays a significant role on timescales much faster than the typical timescales of thrombus growth, and so expansion of upstream vessels is more clearly aligned to the accumulation of blood upstream of the constriction due to the reduced outflow. Using this viscous model, we show that such flows can be captured with a model based only on Stokes flow and observe that as an external pressure is applied to the parent vein, there is a substantial overshoot in blood pressure in the generation of vessels immediately upstream of the occlusion. However, this overshoot diminishes as we increase the perturbation time and eventually becomes negligible. Furthermore, we show that the pressure increases monotonically in generations which are further upstream of the occlusion without exhibiting any overshoot.

In summary, this thesis develops a general framework for building realistic arterial and venous networks from patient images to study the retinal micro-circulation, which can be easily extended to bigger networks. Moreover, at the end, we use our three-generation arterio-venous model to study the real patient data extracted directly from images. Hence, our model does not predict a clear mechanism which might lead to a localised increase in blood pressure, which could drive vessel rupture and help explain retinal haemorrhage in the vessel upstream of the constriction. Further work is therefore necessary to refine the model to robustly predict the onset of retinal haemorrhage.

This research has been presented at several national and international conferences, including the British Applied Mathematics Colloquium (BAMC), the UK Fluids Conference, and the Soft-Mech International Tissue Workshop. Manuscripts are currently in preparation for submission to relevant journals.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Additional Information: Supported by funding from the College of Science and Engineering, University of Glasgow, EPSRC and SofTMech Glasgow.
• Subjects: Q Science > QA Mathematics; R Medicine > R Medicine (General)
• Colleges/Schools: College of Science and Engineering > School of Mathematics and Statistics
• Funder's Name: Engineering and Physical Sciences Research Council (EPSRC), SofTMech Glasgow
• Supervisor's Name: Stewart, Professor Peter, Gao, Dr. Hao and Husmeier, Professor Dirk
• Date of Award: 2026
• Depositing User: Theses Team
• Unique ID: glathesis:2026-85968
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 26 May 2026 13:40
• Last Modified: 27 May 2026 13:08
• Thesis DOI: 10.5525/gla.thesis.85968
• URI: https://theses.gla.ac.uk/id/eprint/85968